def main(config):
# 读取在 DATA_FOR_CORPUS_PATH 目录下的全部xls格式文件
file_names = os.listdir(config["DATA_FOR_CORPUS_PATH"])
chat = []
for file in file_names:
if ".xls" in file:
print(file)
data = pd.read_excel(config["DATA_FOR_CORPUS_PATH"] + "/" + file)
chat += list(data['对话内容'])
# 得到训练语料
fin = open(config["corpus_file"], 'w', encoding='utf-8')
for t in range(len(chat)):
word_segment = []
curr_chat = str.lower(str(chat[t]))
sentence_segment = [i for i in curr_chat.split('\r\n') if i]
if len(sentence_segment) > 3:
for sentence in sentence_segment[:-1]:
if "访客" not in sentence and "客服人员说:" not in sentence:
if "欢迎使用中国银联网络客服" not in sentence:
word_segment += [
word for word in utils.segment(
utils.remove_url(sentence)) if word != ' '
]
if word_segment:
fin.write(' '.join(word_segment) + '\n')
fin.close()
def add_frame(self, image):
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if self.__using_stabilization:
mask = utils.stabilize(foundFace, self.__no_of_frames + 1, image,
face, mask)
hand, hand_contour = utils.get_my_hand(mask, True)
hand_pose, direction = 'None', 'None'
if hand_contour is not None:
motion_detected = self.__motion.get_hand_motion(hand_contour)
else:
return
if not motion_detected:
hand_pose = recognise_hand_pose(
hand,
directly_from_hand=True,
model_path='Models/silatra_gesture_signs.sav')
else:
direction = motion_detected
self.__observations.append((hand_pose, direction))
self.__no_of_frames += 1
def bits(self, s):
"""Compute the bits of entropy in a string.
If the string uses characters missing from the language model,
__init__(default_bits) will be used for that element instead.
Args:
s: String to compute the entropy for
"""
total_bits = 0
for context, value in utils.all_but_the_last(utils.segment(s)):
total_bits += self.get(context).bits(value, self.__default_bits)
return total_bits
def compile(f, n=3):
"""Compile the raw password database.
Builds a histogram of n tuples provided by f.
Args:
f: File containing raw password data
n: tuple length (defaults to 3)
Returns:
LanguageModel
"""
language_model = LanguageModel()
for line in list(f):
language_model.extend(utils.segment(line.strip(), n))
return language_model
def recognise_hand_pose(image,
directly_from_hand=False,
model_path='Models/silatra_digits_and_letters.sav',
using_stabilization=False,
no_of_frames=1):
'''
### SiLaTra Hand Pose Recognition
Provides classification for input hand pose image.
Inputs: (a) Mandatory Parameter - Image for which Hand Pose Classification is to be performed.
(b) Optional Parameters (Use them only if you understand them):
(1) directly_from_hand - boolean - Set this to true if you are passing already cropped hand region in `image` parameter.
(2) model_path - String - If an alternate model is to be used, pass the path of its .sav file.
(3) using_stabilization - boolean - If you intend to use Object stabilization, set this to True. Only use this option if you are classifying hand poses from a continuous feed, else its useless.
(4) no_of_frames - Integer - ONLY TO BE USED IF using_stabilization IS True, pass the number of the frame from the continuous feed you are processing.
'''
import pickle
from sklearn.neighbors import KNeighborsClassifier
if not directly_from_hand:
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if using_stabilization:
mask = utils.stabilize(foundFace, no_of_frames, image, face, mask)
hand = utils.get_my_hand(mask)
if hand is False: return 'No hand pose in image'
features = utils.extract_features(hand)
else:
features = utils.extract_features(image)
classifier = pickle.load(open(model_path, 'rb'))
hand_pose = classifier.predict([features])[0]
return hand_pose
def segment():
raise NotImplementedError()
series_instance_uid = request.args.get('series_instance_uid')
model_id = request.args.get('model_id')
foo = request.args.get('foo',None)
task_id = request.args.get('task_id',None)
return_type = request.args.get('return_type',"html")
if task_id is None:
task = utils.segment(series_instance_uid,model_id,foo=foo)
print(series_instance_uid,task_id,task.ready())
return redirect(url_for('segment',
series_instance_uid = series_instance_uid,
model_id = model_id,
foo = foo,
task_id = task.id,
))
task = utils.celery.AsyncResult(task_id)
print(series_instance_uid,task_id,task.ready())
response = {'ready': task.ready(),'task_id':task_id}
# for debugging
if return_type == "html":
if task.ready() is True:
return render_template("show_results.html",results=response,)
else: # for learning redirect html
return render_template("loading.html",task_id=task_id)
else:
if task.ready() is True:
try:
response["result"]=task.get()
except:
return jsonify(response)
return jsonify(response)
else:
return jsonify(response)
def recognise_hand_pose(image, directly_from_hand=False, model_path='Models/digits_and_letters.sav', using_stabilization=False, no_of_frames=1):
import pickle
from sklearn.neighbors import KNeighborsClassifier
if not directly_from_hand:
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if using_stabilization: mask = utils.stabilize(foundFace, no_of_frames, image, face, mask)
hand = utils.get_my_hand(mask)
if hand is False: return 'No hand pose in image'
features = utils.extract_features(hand)
else: features = utils.extract_features(image)
classifier = pickle.load(open(model_path, 'rb'))
hand_pose = classifier.predict([features])[0]
return hand_pose
def main(config):
print("\n正在进行第一次聚类...\n")
# 读取词向量模型
model = gensim.models.Word2Vec.load(config["w2v_model_file"])
# 读取数据
data = pd.read_excel(config["cluster_excel_file"])
# 得到句子平均词向量
w2v_dim = model.vector_size
print("词向量模型的维度为", w2v_dim)
# 样本的特征矩阵，一列为一个样本
feature = np.zeros([len(data), w2v_dim])
split_word = []
for i in range(len(data)):
word_list = utils.segment(data.iat[i, 5])
feature[i, :], select_word = get_mean_vec(word_list, model, w2v_dim)
split_word.append(' '.join(select_word))
# 只保留分词不为空的数据
bool_idx = [
True if split_word[i] else False for i in range(len(split_word))
]
num_idx = [i for i in range(len(split_word)) if split_word[i]]
data = data[bool_idx]
split_word = np.array(split_word)[num_idx]
feature = feature[num_idx, :]
# 利用sk-learn中的K-means函数进行聚类
y_pred = KMeans(n_clusters=config["n_cluster_first"]).fit_predict(feature)
data["sublabel"] = y_pred
data["cutwords"] = split_word
data = data.sort_values(["sublabel"])
# 保存数据
data.to_excel(config["cluster_excel_file"], index=False)
print("\n第一次聚类完成！\n")
op1 = "QUIT\r\n"
client.send(op1.encode('ascii'))
break
data = client.recv(size,socket.MSG_WAITALL) #Reference: https://www.binarytides.com/receive-full-data-with-the-recv-socket-function-in-python/
# Instead of storing the image as mentioned in the 1st reference: https://stackoverflow.com/a/23312964/5370202
# we can directly convert it to Opencv Mat format
# Reference: https://stackoverflow.com/a/17170855/5370202
nparr = np.fromstring(data, np.uint8)
img_np = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
img_np = imutils.rotate_bound(img_np,90)
img_np = cv2.resize(img_np,(0,0), fx=0.7, fy=0.7)
mask1 = utils.segment(img_np)
gray = cv2.cvtColor(img_np, cv2.COLOR_BGR2GRAY)
# detect faces in the grayscale image
haar_cascade_face = cv2.CascadeClassifier('/home/aarav/Desktop/MajorProject/Models/haarcascade_frontalface_default.xml')
rects = haar_cascade_face.detectMultiScale(gray, scaleFactor = 1.2, minNeighbors = 5);
maxArea1 = 0
faceRect = -1
foundFace = False
for (x,y,w,h) in rects:
if w*h > maxArea1:
maxArea1 = w*h
faceRect = (x,y,w,h)
def HMM(sequence, Dic, ProDic):
lastmaxpro, maxpre = Forward(sequence, Dic=Dic, ProDic=ProDic)
tag = Backward(lastmaxpro, maxpre)
res = segment(sequence, tag)
return tag, res
from utils import segment
import cv2, numpy as np
print('Press q/Q to quit')
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, 30)
lower = np.array([0, 140, 60], np.uint8)
upper = np.array([255, 180, 127], np.uint8)
while True:
_, frame = cap.read()
mask = segment(frame)
'''
Normal segmentation using YCRCB
mask = cv2.cvtColor(frame, cv2.COLOR_BGR2YCR_CB)
mask = cv2.inRange(mask, lower, upper)
'''
cv2.imshow('Original', frame)
cv2.imshow('Segmentation mask', mask)
key = cv2.waitKey(1)
if key == ord('q') or key == ord('Q'): break
cv2.destroyAllWindows()
cap.release()
contour_start=False
lower = np.array([0,147,60],np.uint8)
upper = np.array([255,180,127],np.uint8)
grid = (20,20)
print('gaf0 gaf1')
while(1):
start_time = time.time()
_, frame = cap.read()
x,y,w,h = 100,100,300,300
cv2.rectangle(frame, (x,y), (x+w,y+h), (0,255,0), thickness=2)
roi = frame[y:y+h,x:x+w]
mask = segment(roi, lower, upper)
if not contour_start:
cv2.imshow('Skin segmentation using YCrCb mask',mask)
cv2.imshow('Original',frame)
else:
try:
_,thresh = cv2.threshold(mask,127,255,0)
hand = get_my_hand(thresh)
cv2.imshow('Your hand',hand)
cv2.imshow('Original',frame)
features = extract_features(hand, grid)
predictions = classifier.predict_proba([features]).tolist()[0]
for prob in predictions: print('%.2f'%(prob),end=' ')
classifier = KNeighborsClassifier(n_neighbors=3)
classifier.fit(X, Y)
lower = np.array([0, 147, 60], np.uint8)
upper = np.array([255, 180, 127], np.uint8)
cap = cv2.VideoCapture('good afternoon.avi')
obs = []
frame_no, grid = 1, (20, 20)
print('Frame: [gaf0, gaf1]')
while (1):
try:
_, frame = cap.read()
mask = segment(frame, lower, upper)
_, thresh = cv2.threshold(mask, 127, 255, 0)
hand = get_my_hand(thresh)
features = extract_features(hand, grid)
pred = classifier.predict([features])
print('%5d' % (frame_no), end=': ')
print(pred)
pred = pred.tolist()
obs.append(pred)
frame_no += 1
except:
break
cap.release()
print(obs)
states = ('gaf0', 'gaf1')
payload = fread.readline()
if (payload == '\r\n' or payload == '\n'
or payload == '\r'):
continue
if (not payload):
break
payload = payload.strip()
payloads.append(payload)
y.append(classes[name])
y = np.array(y)
np.save(y_train_dir, y)
for payload in payloads:
tempseg = segment(payload)
if (tempseg == []):
print(payload)
payloads_seged.append(tempseg)
lens.append(len(tempseg))
with open(lens_dir, 'wb') as f:
pickle.dump(lens, f)
model = Word2Vec(payloads_seged,
size=embedding_size,
iter=iter_num,
sg=1,
min_count=min_num,
max_vocab_size=max_voc)
import cv2, numpy as np
from utils import segment
lower = np.array([0, 147, 60], np.uint8)
upper = np.array([255, 180, 127], np.uint8)
cap = cv2.VideoCapture(0)
fourcc = cv2.VideoWriter_fourcc(*'DIVX')
out = cv2.VideoWriter('test2.avi', fourcc, 10, (300, 300))
start_rec = False
while (1):
_, frame = cap.read()
mask, _, _ = segment(frame, lower, upper)
x, y, w, h = 100, 100, 300, 300
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), thickness=2)
roi = frame[y:y + h, x:x + w]
cv2.imshow('You', frame)
if not start_rec: cv2.imshow('Segmented', mask)
else: out.write(roi)
k = cv2.waitKey(50)
if k == ord('q'): break
elif k == ord('s'):
start_rec = not start_rec
cv2.destroyAllWindows()
cap.release()
out.release()